opinion-leader-depth-analysis

name: opinion-leader-depth-analysis description: Systematic depth analysis framework for opinion leaders tracking research completion across 4 layers - L1 Profile, L2 Timeline, L3 Thought Analysis, L4 Ongoing Monitoring version: 1.0.0 metadata: hermes: tags: [wiki, research, depth-analysis, opinion-leaders, quality-control] homepage: https://github.com/kzinmr/ai-topics prerequisites:

skills: [blog-author-thought-analysis, wiki-entity-upgrade, cross-leader-synthesis]

Opinion Leader Depth Analysis Framework

Purpose

Track research completion depth for each opinion leader in the wiki. Ensures consistent quality and prevents duplicate effort by making "what's done" and "what's next" explicit in frontmatter.

The 4 Layers

L1: Basic Profile

• Name, handles, primary platforms
• Blog URL, RSS feed, social accounts
• Current role/company/affiliation
• Brief bio (1-2 paragraphs)

L2: Timeline & Works

• Chronological activity timeline (date x event)
• Major projects with descriptions
• Published articles/blog posts/videos list
• Key milestones and inflection points

L3: Thought Analysis (Core Differentiator)

• Core Philosophy: 3-5 central ideological positions
• Direct Quotes: Verbatim quotes with source URLs (>30% of analysis)
• Thought Evolution: How positions changed over time
• Key Questions: What they are currently exploring
• Conceptual Frameworks: Coined terms, mental models
• Cross-References: Links to other thinkers, concepts, wiki pages
• Citations: Every claim has a source URL

L4: Ongoing Monitoring

• RSS feed registered in blogwatcher
• Social/X account tracked
• Auto-alerts for new posts/tweets
• Periodic review schedule

Frontmatter Structure

research_depth:
  L1_profile: done
  L2_timeline: done
  L3_thought_analysis: done
  L4_monitoring: active
  last_deep_dive: 2026-04-10
  next_review: 2026-05-10
  confidence_score: 0.85
  sources_analyzed:
    - blog: 12 articles
    - x_posts: 50+ tweets
    - podcasts: 2 episodes
    - papers: 1

Quality Checklist for L3 Completion

1. Direct Quote Ratio: At least 30% of thought analysis content should be direct quotes
2. Source Completeness: Every major claim has a URL citation
3. Evolution Tracking: At least 2 time periods compared
4. Framework Extraction: At least 2 named concepts/frameworks identified
5. Cross-Reference Density: At least 3 wiki page links

Workflow

For New Opinion Leaders

1. Check ~/.hermes/scripts/blog_authors.json or ~/x-accounts.yaml for pre-scraped data
2. Run L1: Create basic entity page
3. Run L2: Add timeline and works
4. Run L3: Extract thoughts, quotes, frameworks (use blog-author-thought-analysis skill)
5. Run L4: Set up monitoring

For Existing Pages

1. Read current page and check frontmatter
2. Identify which layers are missing
3. Fill gaps using blog posts, tweets, podcasts
4. Update frontmatter with research_depth
5. Commit: cd ~/ai-topics && git add wiki/ && git commit -m "wiki: depth-analysis <person> L1-L4" && git push

Priority Order

1. Core AI/Agent thought leaders (Karpathy, Willison, Antirez already done)
2. Harness/Infrastructure leaders (Ryan Lopopolo, Boris Cherny)
3. Platform/Tool leaders (Addy Osmani, Simon Willison)
4. Emerging voices (gm8xx8, Daniel Han)

Page Naming Convention

• Use full domain name: simon-willison.md (not simonw.md)
• If duplicate pages exist (e.g., simonw.md + simon-willison.md), merge into canonical name and delete the duplicate
• Check for duplicates before declaring L3 incomplete

Pitfalls

• Don't paraphrase quotes: Use exact wording for citations
• Don't create duplicate pages: Check existing entity pages first
• Don't skip L3: This is the core value - philosophy extraction
• Don't leave stale monitoring: Update L4 status when feeds change
• L3 auto-check must not rely on specific header names: Many completed pages use ## Core Ideas, ## Philosophy, ## Thought Framework etc. Check for substance, not exact header match
• Do NOT track "which sources have been read" as a binary flag: the same source yields different insights at different analytical depths. Track depth level and last_review date instead.
• web_extract timeouts on long pages: Gwern.net and similar long-form essays (30k+ chars) will cause web_extract to time out. Use browser_navigate + browser_console JS DOM extraction instead. See references/long-page-extraction.md for the pattern.